We present maps of jovian cloud properties derived from images taken simult
aneously by the Galileo solid state imaging system (SSI) and the near-infra
red mapping spectrometer (NIMS) at 26 visible and near-infrared wavelengths
, ranging from 0.41 to 5.2 mum. Three regions-the Great Red Spot (GRS), a 5
-micron Hot Spot, and one of the White ovals-were studied, We perform a pri
ncipal component analysis (PCA) on the multispectral images. The principal
components (PCs), also known as empirical orthogonal functions, depend only
on wavelength, The first PC is that spectral function which, when multipli
ed by an optimally chosen number (amplitude factor) at each pixel location
and subtracted from the spectrum there, minimizes the variance for the imag
e as a whole, Succeeding PCs minimize the residual variance after the earli
er PCs have been subtracted off. We find that the pixel-to-pixel variations
at the different wavelengths are highly correlated, such that the first th
ree PCs explain 91% of the variance in the spectra, Further, one can estima
te the amplitudes of the first two PCs using only the four SSI wavelengths
and still explain 62% of the variance of the entire spectrum. This can be a
n advantage when trying to classify features that are resolved in the SSI i
mages but not in the NIMS images. The first PC in all three regions shows n
egative correlation between 5 mum emission and reflected solar light in bot
h atmospheric windows and the methane and ammonia absorption bands. Thus mo
st of the bright, optically thick clouds blocking thermal emission are also
extended vertically to the upper troposphere, The first PC at the GRS show
s a negative correlation between the violet and all other bands except 5 mu
m, for which the correlation is positive, Thus in the GRS there is a red ch
romophore (absorbing in the violet, reflecting at longer wavelengths) which
is associated with clouds that block 5-mum emission, There is no such corr
elation at the hot spot and white oval regions and therefore no chromophore
associated with clouds. The second PC shows a positive correlation between
the depth of the methane and ammonia absorption bands and brightness at ot
her visible and near-IR wavelengths; there is also a negative correlation b
etween these quantities and 5-mum emission. Thus some of the bright, optica
lly thick clouds blocking thermal emission are deep and do not extend verti
cally to the upper troposphere. A color image composed using the first thre
e PCs shows areas of unusual spectra, which appear in distinct colors. An e
xample is the small convective stormlike cloud to the northwest of the GRS,
This cloud is highly reflective at long wavelengths (4 mum) and might indi
cate unusually large particles. (C) 2001 Academic Press.